1. Amirkabir University of Technology, Computer Engineering Faculty, Intelligent Systems Laboratory,Requirements Engineering Course, Dr. Abdollahzadeh 1 Effectiveness of Elicitation Techniques in Distributed Requirements Engineering Abbas Rasoolzadegan

2. Amirkabir University of Technology, Computer Engineering Faculty, Intelligent Systems Laboratory,Requirements Engineering Course, Dr. Abdollahzadeh 2 Introduction Software quality is often reflective of the quality and the software development process of the organization. Most software process models include requirements analysis activities The ultimate quality of the delivered software depends on the requirements upon which the system has been build The later in the software lifecycle a defect is discovered the more expensive it is to rectify

3. Amirkabir University of Technology, Computer Engineering Faculty, Intelligent Systems Laboratory,Requirements Engineering Course, Dr. Abdollahzadeh 3 Introduction (Cont.) Reasons for distributed RE: –Software development organizations are often geographically distributed from their customers and end users –High travel costs –Local availability of skilled technical staff –Face-to-face meetings with end-users are not realistic

4. Introduction (Cont.) Goals of this research –Identify what factors led software engineers to write high quality SRS documents –Evaluate the effectiveness of the “software for collaboration” used to enable distributed requirements engineering –Assess the effectiveness of various requirements elicitation techniques when used in the distributed mode

5. Background Distributed software development projects have become practical because of –Technological improvements in communications structure, bandwidth, performance, –Project members are unwilling to travel –Lack of skilled worker in a geographical area –High travel and relocations costs –Physical location of specialized hardware

6. Background (Cont.) Advantages of email for supporting telecommuting and asynchronous collaboration –Broadcasting capability –Management of communication –Access to information resources –Low communication cost –File transfer –Temporal and spatial flexibility The requirements elicitation techniques which need lower bandwidth channels, work very well in a distributed interaction

7. Empirical Study of Distributed requirements Engineering Conditions –Groups of computer science graduate students from Virginia Tech, role-played as participants in RE effort –All group interaction in the study was distributed Supported by a set of groupware tools that enabled both synchronous and asynchronous collaboration –The customer and engineer participants never met face-to-face to perform any negotiations and discussion related to the project –Problem for research in RE A meeting scheduler system –Two-page requirements document describing the scheduling problem

8. Project Overview A set of collaborative tools –Centra Symposium Supports real time virtual meetings –MOOsburg Facilitate file sharing and informal impromptu meetings –Email File sharing and asynchronous discussions

9. Project Overview (Cont.) The requirements elicitation techniques were used in the project teams –Question and answer method –Customer interviews –Brainstorming and idea reduction –Storyboards –Prototyping –Questionnaires –Use cases –Requirements management

10. Process assessment and results Evaluation methods –Using surveys to determine the software and requirements engineering experience level of participants –Collecting survey data and observations at each planned virtual meeting Meeting sessions were recorded At the conclusion of the project, participants completed an extensive online survey Group spaces in MOOsburg were examined thoroughly and artifacts were archived All email communication between customers and software engineers was monitored and the messages were examined

11. Process assessment and results (Cont.) Four different metrics were applied to evaluate the overall quality of the SRS documents –SRS document grade –Measurement of requirements evolution –Requirement errors –Original requirements supported

12. Process assessment and results (Cont.) Group performance scores: High Performance Groups Reduced Performance Groups Group 179.34%Group 469.11% Group 277.32%Group 666.85% Group 376.44% Group 575.96%

13. Process assessment and results (Cont.) A weak positive trend was seen between ratings of customer participation and overall SRS quality Requirements engineering experience has positive impact on SRS quality A weak positive relationship was seen between overall SRS quality and average perceived peer participation per group A marginally significant negative relationship was observed between requirements elicitation technique effectiveness (Prototyping & Questionnaires) and overall SRS quality

14. Elicitation Method Effectiveness

15. Amirkabir University of Technology, Computer Engineering Faculty, Intelligent Systems Laboratory,Requirements Engineering Course, Dr. Abdollahzadeh 15 Conclusions Distributed requirements engineering is more effective when stakeholders participate actively in synchronous activities of the requirements process Groups who obtained adequate requirements information from the planned virtual sessions had better success writing high quality SRS documents

16. Future work Further empirical studies with more groups Contrast group distributed requirements teams with a control condition of co- located face-to-face meetings What if we had a control group of engineers that only could perform requirements analysis with asynchronous techniques?

17. Reference Wesley James Lioyd, Mary Beth Rosson, James D. Arthur, “Effectiveness of Elicitation in Distributed Requirements Engineering”, Proceedings of the IEEE Joint international Conference on Requirements Engineering, 2002.